Classifications

• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/22—Read-write [R-W] timing or clocking circuits; Read-write [R-W] control signal generators or management 
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
  • G11C7/1051—Data output circuits, e.g. read-out amplifiers, data output buffers, data output registers, data output level conversion circuits
  • G11C7/1066—Output synchronization
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
  • G11C7/1051—Data output circuits, e.g. read-out amplifiers, data output buffers, data output registers, data output level conversion circuits
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
  • G11C7/1075—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers for multiport memories each having random access ports and serial ports, e.g. video RAM
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
  • G11C7/1078—Data input circuits, e.g. write amplifiers, data input buffers, data input registers, data input level conversion circuits
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
  • G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
  • G11C7/1078—Data input circuits, e.g. write amplifiers, data input buffers, data input registers, data input level conversion circuits
  • G11C7/1093—Input synchronization

Definitions

• the present inventions relate to integrated circuit chips and, more particularly, to daisy chained chips including a strobe signal and clock signal with a fixed phase relationship.
• a multi-drop bus is used to connect more than two chips to a set of conductors.
• Dynamic Random Access Memory (DRAM) devices have commonly been used with a multi-drop bus.
• the multi-drop busses can have electrical discontinuities that generate distortion and limit the bandwidth over the bus channel.
• DRAM Dynamic Random Access Memory
• a clock is distributed to each chip.
• the clock is usually distributed in a similar multi-drop fashion, limiting the bandwidth of the clock channel.
• One way to increase bandwidth of the clock channel is to drive multiple copies of the clock to a memory module (such as a dual in line memory module (DIMM)) to limit loading.
• DIMM dual in line memory module
• Another approach is to put a phase locked loop (PLL) component on the DIMM and use this to drive a clock to the DRAM devices.
• PLL phase locked loop
• these approaches may add cost and power consumption to the system.
• point-to-point interconnects have been used to reduce loading and avoid electrical discontinuities. Further, daisy chaine
• FIG. 1 is a schematic block diagram representation of a system according to some embodiments of the inventions.
• FIG. 2 is a schematic block diagram representation of some details of a chip of FIG. 1 according to some embodiments of the inventions.
• FIGS. 3 and 4 are examples of timing diagrams.
• FIG. 5 is an alternative to the details of FIG. 2 according to some embodiments of the inventions.
• FIGS. 6 and 7 are block diagram representations of systems according to some embodiments of the inventions.
• FIGS. 1 is a schematic block diagram representation of a system according to some embodiments of the inventions.
• FIG. 2 is a schematic block diagram representation of some details of a chip of FIG. 1 according to some embodiments of the inventions.
• FIGS. 3 and 4 are examples of timing diagrams.
• FIG. 5 is an alternative to the details of FIG. 2 according to some embodiments of the inventions.
• FIGS. 6 and 7 are block diagram representations of systems according to some embodiments of the inventions.
• FIGS. 1 is
• FIG. 8 and 9 are block diagram representations of a system including memory modules according to some embodiments of the inventions.
• FIG. 10 is a schematic representation of details of transmitters and receivers according to some embodiments of the inventions.
• FIG. 11 is a schematic representation of details a receiver according to some embodiments of the inventions.
• a daisy chained, point to point system includes chips 20, 30, and 40.
• chip 20 may include a memory controller or be a memory device and chips 30 and 40 may be memory devices, although the chips may be other than memory controllers or memory devices.
• Chips 20 and 30 communicate with each other through at least data signals on conductors 50 and a first external strobe signal on conductor 56.
• chips 30 and 40 communicate with each other through at least data signals on conductors 70 and a second external strobe signal on conductor 76.
• An external clock signal (elk) is passed from conductors) 62 to chip 30 to conductors) 82.
• the external clock signal may be the same clock signal as is used by chips 20 and/or 40 or a different clock signal.
• the data signal may be a traditional data signal with address and command signals being supplied separately, or the data signal may be a more generic signal that includes traditional data as well as address and/or command information time multiplexed or packetized with the traditional data signal.
• a first received strobe signal is created in response to the first external strobe signal from chip 20 to chip 30 on conductor 56.
• a second received strobe signal is created in response to the second external strobe signal from chip 40 to chip 30 on conductor 76.
• the first received strobe signal is used to latch data from chip 20 to chip 30, and the second received strobe signal is used to latch data from chip 40 to chip 30.
• An internal clock signal for use by chip 30 is generated from the external clock signal on conductors) 62.
• the phase relationships between the first external strobe signal on conductor 56, the second external strobe signal on conductor 76, and the external clock signal on conductor(s) 62 are arbitrary. Accordingly, the internal clock signal can have a fixed phase relationship with either the first received strobe signal or the second received strobe signal, but not with both the first and second received strobe signals.
• the received strobe signal with the fixed phase relationship is referred to herein as the fixed phase received strobe signal.
• One way to accomplish the fixed phase relationship is to alter the phase of the internal clock signal.
• the received strobe signal that does not have the fixed phase relationship is referred to herein as the arbitrary phase received strobe signal.
• the term low latency data paths refers to those data paths in which data is latched by the fixed phase received strobe signal or the internal clock, but not by the arbitrary phase strobe signal. It is called a low latency data path because the internal clock phase is fixed to the fixed received strobe signal in such a way as to provide low latency to these paths.
• the term arbitrary latency data paths refers to those data paths in which data is latched by the arbitrary phase strobe signal.
• the arbitrary latency data paths may have greater latency than the low latency data paths because the data may wait in a latch longer than is needed while waiting for the internal clock to be received.
• the latency of the arbitrary latency data paths is greater than or equal to, but not less, than the latency of the low latency paths.
• the latency of the arbitrary latency data paths may be slightly lower than that of the low latency data paths, but also can be equal to or greater.
• Chip 30 acts as a repeater between chips 20 and 40 although it may do more than merely repeat data (for example, store data).
• chip 30 repeats (retransmits) at least some data that it receives from chip 20 to chip 40 and repeats at least some data that it receives from chip 40 to chip 20.
• chips 20 and 40 might or might not also be repeaters.
• FIG. 2 provides details of chip 30 according to some embodiments of the inventions, although various other details may be used.
• Receivers 102 and 142 and transmitters 104 and 144 form a first port 92 and receivers 136 and 176 and transmitters 138 and 178 form a second port 94. Note that it is arbitrary which port to label the first port and which to label the second port.
• Transmitters 104 and 138 and receivers 102 and 136 represent multiple data transmitters and receivers on each port (although there might be only one transmitter and receiver per port).
• Conductor 50-1 is one of conductors 50 and conductor 70-1 is one of conductors 70 of FIG.
• FIG. 2 also illustrates conductors 56 and 76 of FIG. 1.
• the external clock signal (elk) of FIG. 1 is a differential clock (Clk and Clk#), but it could be single ended.
• Conductors 62A and 62B and conductor 82A and 82B, which carry the differential external clock signals Clk and Clk#, are examples of conductors) 62 and 82, respectively, of FIG. 1.
• a high bandwidth, low jitter clock can help reduce jitter that could otherwise accumulate at each interface.
• a low voltage swing may also be used.
• receiver 186 and transmitter 188 may include current-mode, differential amplifiers.
• the data and external strobe signals may be full voltage swing or low voltage swing signals.
• the external differential clock signal (Clk, Clk#) is received by receiver 186 which drives the external differential clock signal on conductors 190A and 190B to a delay locked loop (DLL) circuit 196 and transmitter 188.
• Transmitter 188 drives the external differential clock signal to conductors 82A and 82B to be received by another chip or perhaps to be terminated. Buffering provided by receiver 186 may be designed to provide high supply and common mode rejection and very small jitter.
• DLL circuit 196 locks to the buffered external differential clock signal and is selected to lock to a particular phase of the external differential clock signal (for example, 180° (degrees)) and to generate, for example, four equally spaced delays across a complete clock cycle at DLL taps.
• the spaced delays may be fed through the taps to an interpolator 172 to produce, for example, an adjustable delay of 32 steps over 360°.
• Interpolator 172 provides a phase adjusted internal differential clock signal (Clk(I), Clk#(I)) through buffer 162 to a node A (although it could be signal ended).
• the phase adjusted internal differential clock signal at node A is provided to read blocks 116 and 124 and reset synchronizers 158 and 164.
• the phase adjusted internal differential clock signal provides double data rate clocking for data signals passing between array 100 and read blocks 116 and 124, although it could be something else, for example, single data rate or quad-data rate.
• the conductors between array 100 and read blocks 116 and 124 are shown as being unidirectional, but they could be bi-directional.
• the data signals on conductors 50 and 70 and the external strobe signals on conductors 56 and 76 are transmitted as a simultaneous bidirectional (SBD) signals.
• the data and strobe signals may be unidirectional signals or sequential bi-directional signals.
• the data and strobe conductors may be single ended or differential.
• receivers 102 and 136 In response to detecting the data signals on conductors 50-1 and 70-1, receivers 102 and 136 provide first and second received data signals to buffers 108 and 134, which in turn provide the received data signals to write blocks 114 and 126, respectively.
• write block 114 and read block 116 form a first-in first-out (FIFO) circuit 112 in which data is written into the FIFO synchronously with the strobe signal at node B and is read out of the FIFO synchronously with the clock signal at node A.
• FIFO first-in first-out
• write block 126 and read block 124 form a FIFO circuit 122 in which data is written into the FIFO synchronously with the strobe signal at node C and is read out of the FIFO synchronously with the clock signal at node A.
• Reset synchronizers 158 and 164 start a pointer for writing and a pointer for reading at desired places at reset, and may also can keep the latency between writing and reading from being too great or too small (discussed below).
• the external strobe signals on conductors 56 and 76 are in phase with the received data signals on conductors 50-1 and 70-1, respectively.
• the received strobe signals may be delayed by delay circuits 150 and 174 by, for example, 90° and output by buffers 152 and 170 to provide the first and second received strobe signals at nodes B and C, respectively.
• the first and second received strobe signals are also delayed received strobe signals.
• the first and second received strobe signals then have a desired phase relationship with respect to received data signals received by the write blocks 114 and 126 from buffers 108 and 134.
• the received strobe signals at nodes B and C are used by write blocks 114 and 126 to latch the received data signals.
• the received strobe signals may be differential or single ended. Buffers 108 and 152, 134 and 170, and 162 provide desired delays and signal drive. Buffers 108, 152, 134, 170, and 162 maybe carefully matched as desired.
• the first received strobe signal is further delayed by delay circuit 154 under the control of DLL 196 to create a desired phase delay difference between nodes B and A, and hence between writing into and reading from FIFO 112.
• An example of the desired minimum delay is 90°. In the example of FIG.
• delay circuits 150, 154, and 174 are low power consuming slave delay circuits controlled by DLL circuit 196 which acts as a master.
• Interpolator 172 receives the further delayed first received strobe signal from delay circuit 154 and at least part of the phase adjusted internal clock signal from node A and adjusts the phase adjusted internal clock signal to be in phase with the further delayed received strobe signal from delay circuit 154.
• the first received strobe signal is the fixed phase received strobe signal
• the second received strobe signal is the arbitrary phase received strobe signal.
• the circuitry of FIG. 2 could be modified so that just the opposite were the case.
• the delay is 90° which leads to a low latency for those data paths involving FIFO 112.
• the latency in data paths involving both write block 126 and read block 124 is an arbitrary latency because data written into FIFO 122 can take an arbitrary amount of time (within the limits discussed) to be read out.
• FIG. 1 the delay in data paths involving both write block 126 and read block 124 is an arbitrary latency because data written into FIFO 122 can take an arbitrary amount of time (within the limits discussed) to be read out.
• the following data paths are low latency data paths: • the path from conductor 50-1 through receiver 102, buffer 108, write block 114, read block 116, and transmitter 138 to conductor 138; • the path from conductor 50-1 through receiver 102, buffer 108, write block 114, read block 116 to array 100; • the path from array 100 to read block 116 to transmitter 138 to conductor 70-1; and • the path from array 100 to read block 124 to transmitter 104 to conductor 50-1.
• FIG. 3 shows the first received strobe signal at node B being 90° out of phase with the received data signal at write block 114.
• FIG. 4 shows the second received strobe signal at node C being 90° out of phase with the received data signal at write block 126.
• the phase adjusted internal clock signal at node A is 90° out of phase with the received strobe signal at node B.
• the phase adjusted internal clock signal at node A is an arbitrary amount out of phase with the received strobe signal at node C.
• the arbitrary phase difference in FIG. 4 is expressed as a desired minimum phase difference (for example 90°) + X°, where in X may be for example, 360°. In other embodiments, X might be some other amount such as 270° or 450°.
• synchronizers 158 and 164 add phase difference between writing and reading if the phase difference between nodes A and B or A and C gets below the desired minimum amount (in this example 90°) and will remove phase difference if the phase difference gets more than the minimum amount + X (in this example 90° + 360°).
• chip 30 transmits all the data signals that it receives from chip 20 to chip 40. Chip 30 also decides whether the data signals is to be used by chip 30 (for example, stored by chip 30) or are merely to be passed on to chip 40.
• Examples of techniques for deciding whether chip 30 is to use the data include (1) having a particular portion of the data signal currently being received indicate how it is to be used, (2) having a portion of a data signal previously received indicate how a future data signal is to be used, or (3) using a separate chip select signal (not shown). In other embodiments, chip 30 will only pass on only some of the data signals it receives from chip 20 to chip 40. However, passing on only some of the data signals may add significant complexity and/or delay.
• Strobe creation circuitry 156 and 180 creates strobes to be driven by transmitter 144. The strobe creation maybe constant or gated to occur only around the time data signals are to be driven by transmitters 104 or 138.
• FIG. 5 illustrates alternative circuitry to that of FIG. 2. FIG. 5 is like FIG. 2 except that the received strobe signal is provided from receiver 142 to interpolator 172 and there is no delay circuitry 154 between node B and interpolator. Various other alternatives could be used.
• FIG. 6 illustrates a system that is similar to FIG. 1 except that it includes conductors 60 and 80 to carry address and command signals between chips 220, 230, and 240. The conductors are shown as being unidirectional, but could be bidirectional in some embodiments.
• FIG. 7 is similar to FIG. 1 except that the clock signal passes from chip 20 to chip 30 and from chip 30 to chip 40.
• FIG. 8 illustrates chips 20, 30, and 40 of FIG. 1 in a larger daisy chained point to point system where chip 20 is a memory controller that provides data and strobe signals to memory chips Ml 1, M21, M31, and M41 on module 1. Note that, in this example, chip 30 is chip M21 and chip 40 is chip M22.
• Chips Ml 1, M21, M31, and M41 repeat the data signals to memory chips M12, M22, M32, and M42 on module 2, which in turn repeat data signals to memory chips Ml 3, M23, M33, and M43 on module 3, which in turn repeat the data signals to memory chips Ml 4, M24, M34, and M44 on module 4.
• the chips regenerate strobe signals to send to these chips as discussed.
• Clock signals are provided by memory controller 20 and are each passed on to one of the memory modules where they are distributed to memory chips.
• the clock signals could originate other than in the memory controller.
• the modules may be dual in line memory modules (D Ms) or other types of modules.
• FIG. 9 is similar to FIG. 8 except that there are two modules rather than four.
• FIG. 9 is similar to FIG. 8 except that there are two modules rather than four.
• FIGS. 8 and 9 illustrates four chips per module out of convenience for illustration, but there could be more than four memory chips per module and a single clock signal could be distributed to all the chips. In the case of FIG. 9, there could be more than eight memory chips per module. In actual systems, there could be more, less, or the same number of modules as is shown in FIGS. 8 and 9. Rings or other topologies than those shown in FIGS. 8 and 9 may be used.
• memory chips M14, M24, M34, and M44 may be coupled through conductors directly to memory controller 20. Jitter accumulation is a potential problem with the clock daisy-chained through multiple chips. The clock may set the baseline jitter for a chip and retime the data such that the maximum jitter in the chain be held to tolerable levels.
• an SBD signal can be created by transmitter 104 driving a particular signal and receiver 102 monitoring the voltage on conductor 50-1.
• Receiver 102 may subtract the voltage of the outgoing data from the voltage on conductor 50-1 to receive the incoming data signal.
• the amplifier may perform outbound subtraction by use of a pair of selectable reference voltages.
• the state of the output data signal is used to select the appropriate reference voltages, effecting a digital subtraction of the outbound data signal from the data signal present on conductor 50-1.
• the timing of the reference voltage change should be such as to match the output data from transmitter 102.
• Table 1 below shows an example for SBD signaling, where Vcc is the supply voltage for transmitters and receivers of chips 20, 30, and 40 and where a logical high voltage is near Vcc and a logical low voltage is near Vss (ground). This could be at full voltage swing or at a low voltage swing.
• Transmitters 104, 144, 138, and 178 are not restricted to any particular circuitry.
• transmitters such as transmitter 104 may perform as both a termination and a level translator between chip 30 and chip 20 supply voltages.
• the logic power supply (chip 20 Vcc) may be connected to a p-type metal oxide semiconductor field effect transistor (PMOSFET) section of a push-pull transmitter 244 in chip 20, and drive one-half this supply voltage level.
• PMOSFET p-type metal oxide semiconductor field effect transistor
• the transmitters may also serve as terminators, tunable, binary weighted, complementary MOS (CMOS) push-pull transmitters 104 and 244 with linearizing resistors (not shown) may be used.
• CMOS complementary MOS
• the transmitter may implement a slew-rate control function by further segmenting the buffer transistors into four equal subsegments driving in parallel. Delay-line taps may drive these segments to effect a controlled output switching rate, reducing high-frequency content of the channel.
• the Vccs of chips 20 and 30 may be the same or different.
• Receivers 246 and 102 are shown. Low voltage swing may be achieved by, for example, replacing the PMOSFET with an n-type MOSFET (NMOSFET). Note that the transistors do not have to be a metal oxide semiconductor type of FETs or even FETs. The inventions are not restricted to a particular type of receiver. In some embodiments, receivers 102, 142, 136, and 176 may include the details shown in FIG.
• receiver 102 may subtract the voltage of the outgoing data from transmitter 104 from the voltage on conductor 50-1 to receive the incoming data signal; be capable of operating over a wide range of input common mode levels; maintain high common-mode noise rejection; and provide amplification to convert the received signals to CMOS levels.
• receiver 102 may have three stages. The first stage may be based on a pair of complementary differential amplifiers and provide wide common-mode input range and common-mode noise rejection. The amplifier may perform outbound subtraction by use of a pair of selectable references RefHi and RefLo selected by signals HiSel and LoSel.
• the first stage may be further divided into four parallel segments to achieve staged, controlled switching between the two references in order to maintain a closely matched relationship with slew-rate controlled outbound data switching.
• the second stage amplifier may provide gain, common mode restoration, and combine the outputs of the first stage differential amplifiers into a single differential signal.
• the last stage may provide additional gain and convert the differential signal into a single- ended CMOS output at conductor 106.
• There may be various additional circuitry that are not illustrated (for example, electrostatic discharge circuitry).
• An embodiment is an implementation or example of the inventions.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Dram (AREA)
• Electronic Switches (AREA)
• Time-Division Multiplex Systems (AREA)
• Stereo-Broadcasting Methods (AREA)
• Devices For Checking Fares Or Tickets At Control Points (AREA)
• Stabilization Of Oscillater, Synchronisation, Frequency Synthesizers (AREA)
• Synchronisation In Digital Transmission Systems (AREA)
• Optical Communication System (AREA)

Applications Claiming Priority (2)

Publications (2)

Family

ID=34711113

Family Applications (1)

Country Status (9)

Families Citing this family (59)

Family Cites Families (18)

• 2003
  • 2003-12-30 US US10/749,677 patent/US7031221B2/en not_active Expired - Fee Related
• 2004
  • 2004-12-23 DE DE602004023423T patent/DE602004023423D1/de active Active
  • 2004-12-23 KR KR1020067013080A patent/KR100806465B1/ko not_active IP Right Cessation
  • 2004-12-23 EP EP04815495A patent/EP1700308B1/de not_active Not-in-force
  • 2004-12-23 WO PCT/US2004/043426 patent/WO2005066966A1/en not_active Application Discontinuation
  • 2004-12-23 CN CN2004800368450A patent/CN1890754B/zh not_active Expired - Fee Related
  • 2004-12-23 AT AT04815495T patent/ATE444555T1/de not_active IP Right Cessation
  • 2004-12-24 TW TW093140506A patent/TWI269950B/zh not_active IP Right Cessation
• 2006
  • 2006-10-10 HK HK06111135.7A patent/HK1091941A1/xx not_active IP Right Cessation

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events